DFI Admits 5V Vdimm Issue Present

Post by **Apoptosis** » Mon May 30, 2005 7:08 pm

I've been running my DFI SLI-D motherboard with the jumper set to the max without any issues. Topboy from BleedinEdge sent us an e-mail letting us know that their is a problem with the DFI boards and that you should run with the jumper on the 5volt setting because things will blow up. Most of the people who own this board are overclockers and take that risk everyday, so take this with a grain of salt. Wickedld, Capper, infinitevalence, and Apoptosis (Me) all run the 5 volt jumpers for months without anything blowing up!

BleedinEdge's Original Problem Post:

If some of you are not aware, the DFI Nforce 4 Ultra/D/DR SLI models has a jumper setting onboard that allows you to increase the voltage supplied on the Vdimm feed from 3.3 volts to 5 volts. This will give users the option to supply more than 3.2 volts to the ram on their DFI motherboards.

Mr Wong basically informed me that DFI R&D in Taiwan know about this issue and have been testing this jumper setting with many different major ram modules. He told me that the 5 volt setting is causing and i quote "Major Issues".

Mr Wong went on to say that DFI's R&D in Taiwan recommended to him that the 5 volt jumper be set to the default 3.3 volt setting on the board while the problem exsists.

Mr Wong also said "At this time they do not have a fix for this problem. Our R&D has also found that the Vdimm jumper when set to 5 volts can cause some of the motherboard chipsets to blow"

DFI Street's Answer To Memory Blowing Up:

To conclude, there has been no evidence that the DFI NF4 motherboard has been responsible for any memory death. You can choose to not believe this if that is what you want to do. We have done extensive testing and found no evidence that the board kills memory modules though, and that is the truth, no matter how much some of you might want to flame and shout that you 'know for sure' that the board does kill memory.

Anyone else remember my UTT article? ;)

infinitevalence · Post by **infinitevalence** » Mon May 30, 2005 9:07 pm

Well i have been using this mobo since the fisrt month it was released and i have had it set using the 5v jumper. I have had no problems with my memory which has been running at 3.4v non stop since i got the board. As long as you folow the prodecdures layed out in the manual switching from 3.3v to 5v should not be any sort of problem.

LVCapo · Post by **LVCapo** » Mon May 30, 2005 10:03 pm

I wrote the review, and had a working board a couple of months before they hit the street, and had no problem other than a few early board bugs. I have also been using this board in my main system since day one and love it.
As Infinite said, I think the problem lies with people not reading the manual or doing the switch the correct way......its always easier when you do something stupid to complain about the product or blame someone else.

TheGeekMistress · Post by **TheGeekMistress** » Tue Feb 14, 2006 1:16 am

what i've heard, from posting on the street for a while, that it's the 4volt jumper and the problem lays in the mosfet that keeps the current clean. this mofet gets soo hot that it needs a decent heatsink, but DFI did not nor is placing a heatsink on this mosfet. it's so bad that the mosfet can't hold the current properly, overheats, then melts the 4v jumper and with a raw uncontrolled current running thru your ram, it fries the ram out.

the new Expert mobos don't come with the 4v jumper option at all. but those board are having stability issues with many users.

TGM

infinitevalence · Post by **infinitevalence** » Tue Feb 14, 2006 1:31 am

Welcome

TGM

The easy solution to the DFI jumper issue is good cooling like you said. I keep a 3 80mm fans cooling the ram, 5v mosfett, and PWM area. Thanks to that i enjoy extremely stable ram and cpu voltages. And because of all that good cooling i run my ram at 3.6v 24/7 at 260MHz

// god i love my ultra-d sometimes \\

now your sig issue, may be related to the number of lines, try putting it into fewer lines with the same number of CHAR and see if that works

TheGeekMistress · Post by **TheGeekMistress** » Tue Feb 14, 2006 1:45 am

infinitevalence wrote:Welcome TGM

The easy solution to the DFI jumper issue is good cooling like you said. I keep a 3 80mm fans cooling the ram, 5v mosfett, and PWM area. Thanks to that i enjoy extremely stable ram and cpu voltages. And because of all that good cooling i run my ram at 3.6v 24/7 at 260MHz // god i love my ultra-d sometimes \\

now your sig issue, may be related to the number of lines, try putting it into fewer lines with the same number of CHAR and see if that works

Thanx for the welcome

why do people here call it the 5v jumper when everyone at the street calls it the 4v jumper? it only goes up to 4volts say the "support" peeps.

just a little quote here

Most extensive BIOS I have ever used,more tweaking options than you have ever seen.BIOS has the capability to go to 4.0 volts on memory....

TGM

infinitevalence · Post by **infinitevalence** » Tue Feb 14, 2006 2:47 am

i call it the 5v jumper because while it does not go to 5v it is based of the 5v power rail rather than the 3.3v rail which after filtering and such tends to only give up to 2.9v for memory. That is why lots of motherboards are limited to 2.9v with out a voltmod. so hense the 5v moniker.

pastorjay · Post by **pastorjay** » Tue Feb 14, 2006 8:09 am

Yes! Welcome TGM!

Good to have you here at LR!

Post by **Apoptosis** » Tue Feb 14, 2006 8:25 am

Welcome to the forums TheGeekMistress!

To answer you're question on why people here call it the five volt rail I'm not sure. The older DFI boards got the VDIMM voltage from the 3.3V rail and when the newer "high voltage" boards came out they moved over to the 5V rail to get more power. I guess this is why I call it the 5V jumper even though it is limited to 4 by the bios if I recall correctly.

kenc51 · Post by **kenc51** » Tue Feb 14, 2006 8:34 am

TheGeekMistress wrote: the new Expert mobos don't come with the 4v jumper option at all. but those board are having stability issues with many users.

TGM

I have the "expert" board......When they are running they are great...

I'm now having probs with fluctuating Vdimm.....Anything over 2.8v is fine
under this ~2.6v - 2.7v is goes all over the place...
I've tested it with a multimeter and it keeps dropping to 2.55v
I use the G.Skill 2GB kit and it needs 2.6v for me to keep it stable
I've now left the pc unplugged with the battery out and CMOS jumpper set...I'll leave it till I get home tonight, but I thing I'm gonna have to RMA this board..... and yes I've reflashed the bios
I'm thinking of getting either the Asus A8N32-SLI Deluxe or the MSI K8N Diamond Plus or just go back to the "standard" sli-dr......

Welcome TheGeekMistress BTW!!!

TheGeekMistress · Post by **TheGeekMistress** » Wed Feb 15, 2006 4:22 am

w0w, thanx for all the welcomes

for those of you using ram like my BH5, i have a question as to how much voltage you are giving that ram and your cpu?

there are some leaked AMD reported dangers of having a voltage diferential of 1.7~1.9 between the on die memory controller and the ram.

what AMD said "unofficially" is that you need to bump up your cpu voltage to be safe when using high ram voltages.

how i came to the voltage diferential that i did, was to take the lowest allowed voltage that AMD uses, the QnC function. it goes down to 1.1vcore. so if we take the 1.1volt vcore and the highest JEDEC voltage allowed 2.8volts and subtract them, the difference is 1.7volts. thats the basic way and more conservative difference to go by.

this means that if you run a 1.4vcore add 1.7 and your ram voltage should only be allowed to go to 3.1, which happens to be the highest setting that the DFI nf3 250gb goes for the vmem. but if you're running a 1.55vcore then add 1.7 and your highest vmem should be 3.25volts.

this is all in the wake of people having thier ram sticks not working in other mobos or posting in their new DFI mobos. or else, and this is the worst part, is that the memory controller that on die is fried from trying to take a 1.5vcore and a 3.6vmem differnce and it just burns the connecting diodes out.

i have a link around here somewhere from XS's Bigtoe (Tony from OCZ) explaining all this, but i'm a bit too tired to try and search thier forums right now. sorry.

i have 2x256 BH5 and 2x512 TCCD G.Skill and b/c the BH5 requires soo much voltage, i'll be using, only, my TCCD in my new A64 mobo. i was able to get my G.Skill running on my nf2 mobo at 240 2.0-2-2-2-5 with only 2.9volts so i feel good about that.

TGM

Post by **Apoptosis** » Wed Feb 15, 2006 8:32 am

Damn now I know why you have the name "TheGeekMistress"... LOL

Glad to have you on the forums as you know whats going on!

infinitevalence · Post by **infinitevalence** » Wed Feb 15, 2006 10:44 am

/me raises hand

That would be me. I have been running my memory at 3.6v for quite some time now, and the CPU tends to run at 1.5-1.6v depending. Its 1 year since i got my Ultra-D and i have not yet burned out a CPU. Its not that i dont belive them but i think its far more rare they some people suggest.

Bio-Hazard · Post by **Bio-Hazard** » Wed Feb 15, 2006 12:42 pm

I'm also running the DFI LP Ultra D. Although I don't run BH-5 ram all the time, I do run it for extended periods of time at 3.6 volts or higher every now and them and the CPU voltage stays up pretty high also. I have yet to have any problems with the CPU, MoBo or ram, but that's not saying that it couldn't happen.

TheGeekMistress · Post by **TheGeekMistress** » Wed Feb 15, 2006 7:28 pm

infinitevalence wrote:/me raises hand

That would be me. I have been running my memory at 3.6v for quite some time now, and the CPU tends to run at 1.5-1.6v depending. Its 1 year since i got my Ultra-D and i have not yet burned out a CPU. Its not that i dont belive them but i think its far more rare they some people suggest.

well, a vcore of 1.5+ 1.7 = 3.2vmem and a less conservative viewpoint of voltage diferential of 2.9-1.1=1.8 so 1.5cvore + 1.8 diferential = 3.3 ..or as you say you run your voltage upwards to 1.6vcore 1.6vcore+1.8 diferntial = 3.4vmem.

these are just some numbers to play with.

the memory controller burn out length of time also depends on the micron processes of your cpu; the .09µ cores are at most risk. also which REV. memory controller you have ie. CG D0 E3 E4 E5 E6. also, this isn't going to absolutely kill your cpu or memory sticks right away. to start out you'll get less and less overclock on your memory bus oc and you'll start having more crashing of your cpu at speeds that used to be prime stable. or, if the worst case senerio happens to you, you'll have a weak mem controller and your ram will die. this has been a huge problem with the DFI mobos and ram dieing pretty quickly as the DFI mobos can give voltage up pretty high, like you say 3.6vmem.

it all depends on the luck of the draw, we all know that some mem controllers are more stout than others, this goes for the DRams on your mem sticks as well.

the point i was tryingto make, is that thier is confirmed data out there that this is/will happening.

read for yourselves
http://www.xtremesystems.org/forums/sho ... hp?t=79509

TGM